Microparticle Preparation by a Propylene Carbonate Emulsification

International Journal of Pharmaceutics 544 (2018) 213–221

Contents lists available at ScienceDirect

International Journal of Pharmaceutics

journal homepage: www.elsevier.com/locate/ijpharm

Microparticle preparation by a propylene carbonate emulsiﬁcation- T extraction method ⁎ Daris Grizića, , Alf Lamprechta,b a Department of Pharmaceutics, Institute of Pharmacy, University of Bonn, Gerhard-Domagk-Str. 3, 53121 Bonn, Germany b PEPITE (EA4267), University of Burgundy/Franche-Comté, Besançon, France

ARTICLE INFO ABSTRACT

Keywords: The use of various harmful organic solvents for microparticle formulations is still widespread. Here, an alter- Propylene carbonate native low toxicity solvent (propylene carbonate; PC) is proposed for the preparation of poly(lactic-co-glycolic- PLGA acid) (PLGA) microparticles. Based on the classical emulsification-solvent extraction methodology, the use of PC Enhanced solvent extraction offers the unique advantage of an additional solvent extraction step using hydrolytic solvent cleavage during Microparticles microparticle preparation. Spherical, rough-surfaced microparticles were obtained with a volume median diameter range from 20 to 60 µm. The residual PC content has been identified to be the major factor for the solidification hindrance, leading to polymeric Tg shifting due to a plasticizing effect. When applying the enhanced PC extraction step, the residual PC content was lowered from 8.8% to 2.7% and subsequently Tg values shifted from 8.2 to 37.7 °C. Additionally, the hydrolytic solvent cleavage confirmed to have no impact on the PLGA stability. This method presents a significant advancement towards replacing of conventional solvents in the microparticle preparation due to more efficient solvent extraction.

1. Introduction Potentially toxic solvents are needed to dissolve hydrophobic polymers like PLGA or PLA, despite using moderate preparation con- Various pharmaceutical formulations nowadays still rely on the use ditions which are appropriate for sensitive drugs (Bitz and Doelker, of organic solvents. This is particularly true for microparticulate par- 1996). As an alternative, non-toxic solvents could be advantageous enteral formulations intended for controlled drug release of small mo- because they can overcome the safety-related issues. Hence, the use of lecules or protein drugs. The microencapsulation of these substances is non-toxic polymer solvents for multiparticulate systems can be sug- usually based on an emulsification – solvent elimination approach (Ao gested to avoid the issue of a complete residual solvent removal. These et al., 2011; Rosca et al., 2004; Shao et al., 2017). In general, an initial solvents possess a considerable advantage, since they can remain within oil-in-water emulsification step is employed, followed by the elimina- the formulation after preparation of the microparticles due to their low tion of the inner organic phase performed by either extraction or eva- toxicity. Solvents like dimethyl sulfoxide, glycofurol and liquid poly- poration (depending on the vapour pressure of the organic solvent) ethylene glycols have been previously used in this manner (Ali and (Katou et al., 2008; Vay et al., 2012). Lamprecht, 2013; Allhenn and Lamprecht, 2011; Viehof et al., 2013). Different organic solvents are used for the formulation of micro- However, the use of these solvents involves formulation issues such as particulate drug carrier systems (Song et al., 2006). Among the most high viscosity, low drug solubility, potential stability problems, etc. current ones are non-halogented solvents, like ethyl acetate or iso- Previous reports suggested that using ester-type solvents like methyl propanol, but also halogenated solvents like 1,2-dichloromethane. propionate (Kim et al., 2016) and ethyl formate (Sah, 2000), both being However, according to ICH guidelines for residual solvents Q3C(R5), partially water-soluble, can be a good alternative for the production of halogenated solvents possess potential toxic properties belonging to the microparticles, while exhibiting low toxic properties. class II solvents (ICH, 2016). Formulations prepared with class III sol- Here, we propose a new formulation technique based on propylene vents such as acetone or ethanol typically are allowed to contain more carbonate (PC) as an alternative low toxic ester-type organic solvent for “parts per million” residual solvent, but the final removal below the microparticle preparation intended for parenteral administration. PC is permitted threshold after microparticle preparation can be technically a member of cyclic organic carbonates, miscible with most organic challenging (Bitz and Doelker, 1996; Herberger et al., 2003). solvents like acetone, ethanol, chloroform etc. (Fujinaga and Izutsu,

⁎ Corresponding author. E-mail addresses: [email protected] (D. Grizić), [email protected] (A. Lamprecht). https://doi.org/10.1016/j.ijpharm.2018.03.062 Received 12 January 2018; Received in revised form 16 March 2018; Accepted 31 March 2018 Available online 06 April 2018 0378-5173/ © 2018 Elsevier B.V. All rights reserved. ć D. Grizi , A. Lamprecht International Journal of Pharmaceutics 544 (2018) 213–221

1971; Raymond et al., 2009). Also, it is freely miscible with water at 2.2.2. PC hydrolysis tracking concentrations up to 20% (Shaikh and Sivaram, 1996). The ability to The PC hydrolysis tracking was accomplished using thymol blue dissolve a wide range of polymers makes PC an attractive alternative to (TB) as a pH shift indicator which occurs during PC hydrolysis. The commonly used solvents. major analytical drawback for the hydrolysis tracking of PC is the op- However, in the context of alternative safe solvents, plasticization of tical inertness which it exhibits both in UV and VIS region (Fujinaga the polymeric matrix has been identified to be the major issue involved and Izutsu, 1971; Grizić et al., 2016). For this reason, an indirect de- in microparticle design (Jain et al., 2000; Katou et al., 2008; Sah, tection method was employed by using the ability of thymol blue (TB) 1997). This is especially pronounced for water miscible or partially to exhibit pH-dependant color transitions in the regions between miscible solvents like glycofurol or ethyl acetate (Allhenn and pH < 8.0 (yellow) and pH > 9.6 (blue). During PC hydrolysis using Lamprecht, 2011; Sah, 1997). Consequently, solvent-based plasticiza- aqueous sodium hydroxide, ring opening of PC (cyclic ester) occurs, tion is the major hindering factor for microparticle solidification if the which leads to the formation of propylene glycol and sodium hydro- residual solvent quantity is not lowered. gencarbonate. If excess amounts of sodium hydroxide are present, so- In terms of safety considerations, the non-toxicity of PC is under- dium carbonate is formed. For this reason, we evaluated aqueous so- lined in various reports (Beyer et al., 1987; Das et al., 2017; Quintanar- lutions of these potentially forming substances in stoichiometric Guerrero et al., 1996; Sommer et al., 1990). PC undergoes two de- identical concentrations which are formed during the actual micro- gradation pathways either by acid/base-induced hydrolysis (Shaikh and particle preparation using TB and retrieved the respective spectra (Fig. Sivaram, 1996) or enzyme-catalyzed hydrolysis in vivo (Yang et al., S1). The end-point of PC hydrolysis gives a solution with two absorp- 1998). In both cases, cyclic organic carbonates produce carbonic acid tion maxima at 434 nm and 597 nm, respectively. In brief, 5 ml of 2% and 1,2-diols, where the type of the produced diol is dependent on the Na2CO3, 1.5% NaHCO3, 0.15% NaHCO3 and 2% PC were mixed with type of cyclic organic carbonate, confirming the safe degradation of PC 0.05 ml 0.1% ethanolic TB solution and analyzed using a UV–VIS into carbon dioxide and propylene glycol (Clements, 2003). Accord- spectrophotometer (Lambda 12, PerkinElmer UV–Vis spectro- ingly, we were able to enhance the solvent extraction from the poly- photometer, MA, USA), recording their spectra from 400 to 700 nm. meric matrix by the chemical degradation of PC, making PC much more Secondly, the optimal process parameters regarding the hydrolysis of suitable as a polymer solvent compared to non-toxic solvent approaches PC (dropping speed and concentration of sodium hydroxide) which at that have been reported before. the end could affect the stability of the excipients, had to be found. A constant amount of PC (100 mg) and varying concentrations of sodium hydroxide, expressed as the percentage of the maximum stoichiometric 2. Materials and methods amount which is needed for a complete reaction (39.18 mg sodium hydroxide), were used. The analysis was performed using a 1 cm quartz 2.1. Materials cuvette, filled with a mixture of 50 µl 0.1% TB solution and 2 ml 5% PC. Immediately after adding the sodium hydroxide solution, continuous PLGA [Poly(DL-lactide-co-glycolide)] (Resomer® RG 502H) was ob- time-dependent measurements at 434 nm and 597 nm were performed, tained from Boehringer Ingelheim (Germany). Propylene carbonate measuring the absorbance every 2 sec during 30 min. This procedure (PC) was purchased from Merck (Darmstadt, Germany). Glycofurol, was repeated for all sodium hydroxide concentrations. Different con- sodium carbonate, methanesulfonic acid, lactic acid, glycolic acid, so- centrations of sodium hydrogencarbonate (the major product during PC dium hydroxide and hydrochloric acid were obtained from Sigma- hydrolysis) gave different intensities, but always the same intensity Aldrich (Steinheim, Germany). Thymol blue, sodium dihydrogen ratio between the two absorption maxima, which was 0.66. This value phosphate, sodium hydrogen carbonate and sodium sulfate were pur- was the fixed end-point in all further investigations. The influence of chased from Roth (Karlsruhe, Germany). Polysorbate 80 was obtained the sodium hydroxide concentration on the speed of hydrolysis was from Caelo (Hilden, Germany). All other chemicals were of analytical evaluated (the linear relationship is shown in Fig. S2), giving the insight grade. into the needed hydroxide ion concentration which has the shortest residence time in the solution (2 ml of a 2 M sodium hydroxide solution dropped at a dropping speed of 20.0 µl/min and a dropping rate of 2.2. Methods 1 drop/45 s).

2.2.1. In-situ drop to microparticle transformation 2.2.3. Hydrolytic profiling of PC, PLGA and polysorbate 80 The drop to microparticle transformation was microscopically Using the TB-based hydrolysis tracking method, hydrolytic profiles evaluated using a Leica DM 2700 M microscope (Leica Microsystems, of PC, PLGA and polysorbate 80 were evaluated. The hydrolysis of PC Wetzlar, Germany) equipped with a QImagingMicroPublisher 5.0 Real- and polysorbate 80 was evaluated directly in a quartz cuvette by adding Time Viewing camera (QImaging, Surrey, BC, Canada) and recorded 0.24 ml of a 0.002% sodium hydroxide solution into the premixed PC/ using QCapture Suite software. Two experimental approaches have TB and polysorbate 80/TB solutions and immediately measuring the been used: drop transformation during microparticle preparation and color transition at 597 nm over 3 h. For PLGA, 5 mg of the polymer was detailed observation of a single droplet during PC diffusion. Both ob- dispersed in 2 ml of water and mixed with 50 µl 0.1% TB solution. servations were done without enhancement of the PC extraction. For 0.24 ml of a 0.002% sodium hydroxide solution was added and the the first approach, samples were drawn at different time points during suspension was filtrated (0.2 µm) at predetermined time intervals and the preparation and directly observed. The experimental setup of the analyzed at 597 nm also for 3 h. It is important to note that the final second approach consisted of a polystyrene petri dish with a micro- sodium hydroxide concentration for microparticle preparation and for scope glass slide which was mounted on the microscope stand. The glass the hydrolytic profiling were stoichiometric identical (0.00024%). slide was used in order to prevent the instant droplet collapse due to the high affinity of PC for polystyrene. A 1% PLGA in PC solution was 2.2.4. Microparticle preparation prepared and mixed with nile red as a lipophilic stain. 40 ml of a An emulsification – solvent extraction method was employed for the 0.004% polysorbate 80 solution (corresponding to the final polysorbate preparation of all microparticle samples. In brief, 100 mg of PLGA 502H concentration in the extraction medium during microparticle prepara- was dissolved in 10 ml of propylene carbonate, giving a 1% PLGA/PC tion) is added to the petri dish. Using a 1 ml syringe with a 30 G needle, solution. Thereafter, 25 ml of a 0.1% aqueous polysorbate 80 solution a small drop of the nile red stained PLGA – PC solution was introduced was added, leading to a biphasic system. Subsequently, this mixture into the petri dish and recorded during 30 min. was stirred by a propeller stirrer (IKA RW 20 digital, 4-bladed stirrer,

214 ć D. Grizi , A. Lamprecht International Journal of Pharmaceutics 544 (2018) 213–221 shaft size 8 mm × 200 mm, stirrer diameter 35 mm) at 400 rpm for step for stability study) were placed in non-hermetically sealed alu- 2 min, leading to the formation of an o/w emulsion. The formed minum pans and equilibrated at −60 °C for 5 min. Afterwards, all emulsion was immediately added to 500 ml of distilled water which samples were heated from −60 °C to 250 °C at a rate of 10 °C/min. The was kept stirring at 250 rpm for 100 min. In order to improve the mi- samples were again cooled down to −60 °C and after a repeated croparticle solidification, an enhanced PC extraction was integrated in equilibration at −60 °C, the heating cycle was repeated. The results the preparation procedure using hydrolytic treatment of the formed were analyzed using STAReSW 13.0 software. Additionally, micro- emulsion by adding 2 ml of sodium hydroxide (2 M) drop wise using a particle stability was evaluated at three storage conditions: peristaltic HPLC pump (with a dropping speed of 20.0 µl/min and a 25 °C ± 2 °C/60% RH ± 5% RH, 5 °C ± 3 °C and 40 °C ± 2 °C/75% dropping rate of 1 drop/45 s). Additionally, microparticles which were RH ± 5% RH according to ICH guideline Q1A(R2) using a potential Tg not subjected to the enhanced PC extraction step (no addition of sodium shift investigation. hydroxide), were also prepared. The obtained suspension was cen- trifuged at 800 rpm during 3 min, the supernatant removed and the 2.2.10. Quantification of residual PC pellet washed with distilled water. Finally, the microparticles were 10 mg of microparticles were dissolved in 100 µl of glycofurol and collected by filtration and dried in a desiccator overnight. then 900 µl of distilled water were added to the clear solution to pre- cipitate PLGA. After filtration through a 0.2 µm polypropylene mem- 2.2.5. Particle size distribution brane, the clear aqueous filtrate was assayed for PC content as de- Laser diffraction (Helos, Sympatec®, Clausthal, Zellerfeld, Germany) scribed previously (Grizić et al., 2016). The limit of quantification was employed in order to investigate the size change during the drop to (LOQ) for the used analytical method was 3.1 ± 1.4 µg/ml, allowing microparticle transformation, expressed as the volume distribution of the quantification of the residual PC content in the microparticles. All the particles. For this purpose, deionized water (as during microparticle results were expressed as percentage [m/m]. preparation) was used for the analysis, while the optical concentration was maintained at 3%. All samples were analyzed in triplicate. 3. Results

2.2.6. Scanning electron microscopy The microparticle preparation was based on an emulsification – A scanning electron microscope (Hitachi SU3500, Tokyo, Japan) solvent extraction method, where PC along with the dissolved PLGA was used to evaluate the microparticle morphology of all samples, at acted as the inner phase and aqueous polysorbate 80 as the outer phase. 10 kV. Firstly, all microparticle samples were mounted on aluminum After an o/w emulsion was formed, an excess amount of water was supports using double-adhesive tape and gold-coated using a sputter- added leading to solvent extraction. A gradual transformation initiated coater (Polaron SC7640 Sputter Coater, Quorum Technologies Ltd., by droplet shrinkage and finished with an apparently complete solidi- Newhaven, UK). Finally, the samples were placed onto the sample fication was observed at consecutive time points (Fig. 1A–C). holder of the scanning electron microscope and analyzed. Additionally, the drop to microparticle transformation was observed using a single droplet setup in order to assess the detailed inner and 2.2.7. Confocal laser scanning microscopy outer morphology change during microparticle solidification which ® A Nikon Eclipse Ti Al Laser Scanning Confocal Imaging System lasted typically for 15–20 min (Video 1). (Nikon Corporation Inc., Tokyo, Japan) equipped with a modular laser system and an inverted Nikon® microscope was used to analyze the microparticles. The argon laser was run at 488 nm with a pinhole size of 1.5 A.U. In order to investigate the distribution of residual PC, sodium fluorescein was chosen since it dissolves in propylene carbonate, but does not stain PLGA. Samples with pre-stained propylene carbonate were prepared without and with the enhanced PC extraction step and analyzed.

2.2.8. HPLC evaluation of PLGA degradation PLGA degradation was evaluated using HPLC, based on previous findings which showed that the tracing of PLGA monomers gives a good representative image of the overall degradation profile of the polymer (Li et al., 2012). For this purpose, pure PLGA, PLGA mixed with either polysorbate 80 or PC and a microparticle preparation mixture (PLGA, polysorbate 80 and PC) were used. At the end of the hydrolytic treatment, aliquots were withdrawn and analyzed using an Acclaim™ OA, 250 mm × 4 mm, 5 µm (Thermo Scientific) column. The analysis was performed at 30 °C using a flow of 0.6 ml/min of the mobile phase, which consisted of 100 mM Na2SO4 adjusted to pH 2.65 with metha- fi nesulfonic acid. 25 µl of pre- ltered and degassed sample was injected Video 1. Drop to microparticle transformation observation using a single dro- in each run and detected at 210 nm. The obtained calibration linearity plet microscope setup reveals the detailed inner and outer morphology change range was between 5 and 1000 µg/ml (R = 0.9994 for glycolic acid and during microparticle solidification. The observation was done without en- 0.9999 for lactic acid). The mean retention times for glycolic acid and hancement of the PC extraction. Initial experiments which employed lactic acid were 4.5 min and 5.7 min, respectively. drying of the apparently solid microparticles which were not subjected to the enhanced PC extraction step resulted in microparticle coales- 2.2.9. Differential scanning calorimetry cence and aggregation, eventually leading to polymeric film formation DSC examinations were carried out using a Mettler Toledo DSC2 (Fig. 2A). The product was further evaluated in terms of residual PC, instrument (Columbus, OH, U.S.A.), which was calibrated using indium revealing a content of 8.8 ± 0.1%. On the other hand, the application as a standard. All samples (PLGA, PC, PLGA microparticles without the of the enhanced PC extraction step during preparation inhibited the enhanced PC extraction step, PLGA microparticles with the enhanced microparticle aggregation and film formation (Fig. 2B), leading to the PC extraction step and microparticles with the enhanced PC extraction obtainment of dry microparticles. In this case just 2.7 ± 1.3% of

215 ć D. Grizi , A. Lamprecht International Journal of Pharmaceutics 544 (2018) 213–221

Fig. 2. Investigation of the microparticle morphology and coalescence tendency of PLGA microparticles prepared without (A) and with (B) the enhanced PC extraction step. Both samples had an extraction time of 100 min. The scale bar represents 100 µm.

only initiated, allowing the selective hydrolysis of the solvent only. Also, it was observed that polysorbate 80 had a profoundly higher hydrolytic resistance towards the alkaline solution compared to PC, where PC hydrolysis was completed when polysorbate 80 hydrolysis did not even initiate. In addition, the feasibility investigation of the enhanced PC extraction step was finalized by assessing sample aliquots obtained during microparticle preparation for degradation-based acidic monomers. For Fig. 1. Microscopic images of the drop to microparticle transformation process this purpose, pure PLGA, PLGA with the addition of polysorbate 80, during (A) emulsification, (B) solvent extraction after 5 min and (C) solvent PLGA with the addition of PC and a full microparticle composition extraction after 30 min. Sample formulation was done without the enhanced PC (PLGA, polysorbate 80 and PC) were subjected to the enhanced PC extraction step during preparation. The scale bar represents 1000 µm. extraction step as described in the microparticle preparation Section 2.2.4. After the total amount of sodium hydroxide was added drop wise, residual PC was present. Even though both samples had the same ex- the supernatants were analyzed using HPLC (Fig. 5B). Pure PLGA traction time of 100 min, differences in the physical appearance were showed a high amount of degradation products (lactic and glycolic significant. Consequently, residual PC was identified to have a sig- acid) after being treated. PLGA degraded in a lesser extent when just nificant impact on the possibility to obtain a non-aggregated product. polysorbate 80 was present and no PC was added. Finally, PC con- In order to evaluate the internal structure of the microparticles re- taining samples (alone or with polysorbate 80) did not result in PLGA garding a possible residual PC localisation, confocal laser scanning degradation, pointing to the high reactivity of PC compared to PLGA microscopy was used. Using fluorescein, a hydrophilic dye which stains and polysorbate 80. PC while leaving PLGA unstained, a localisation pattern was observed. The influence of the duration of the enhanced PC extraction step on While an almost continuous PC distribution throughout the micro- microparticle solidification was evaluated for periods of 0.5 h, 1.5 h and particle matrix could be observed for the sample which was untreated 4.5 h, respectively. Changes in terms of morphology were identified (Fig. 3A), the enhanced PC extraction led to PC depletion and locali- using SEM (Fig. 6). It could be observed that samples with a longer sation mainly in the interior cavities (Fig. 3B). enhanced PC extraction (≥1.5 h) showed a complete solidification, The transition from emulsion droplet to microparticle during the while the shorter lasting extraction (0.5 h) resulted in solidified and un- preparation step was distinctly slower in absence of the enhanced PC solidified microparticles (which can also be noticed by the decreased extraction step and led additionally to an increased droplet diameter number of solid microparticles in the 0.5 h sample). Additionally, these until sufficient particle solidification took place (Fig. 4A). Oppositely, samples were also tested for PLGA degradation products and the results the application of the enhanced PC extraction step resulted in the final showed that no degradation products were present, pointing to the size distribution already at the first data point (Fig. 4B). integrity of the matrix even at longer extraction times (data not shown). In order to exclude the possibility that the enhanced PC extraction The plasticizing effect of residual PC which has a strong impact on step will accidentally degrade other compounds than PC, namely PLGA PLGA was assessed by DSC measurements, focusing on the Tg shift. Due or polysorbate 80, their respective hydrolysis was experimentally as- to the fact that PC has a low Tg (−114.6 °C), it represents a potential sessed (Fig. 5A). PC hydrolysis was completed when PLGA hydrolysis plasticizing agent. The absence of the enhanced PC extraction step

216 ć D. Grizi , A. Lamprecht International Journal of Pharmaceutics 544 (2018) 213–221

Fig. 4. Droplet size measurements in the o/w emulsion over time (A) without and (B) with the enhanced PC extraction step.

Fig. 3. Localisation of residual PC in the microparticle matrix of microparticles without (A) and with (B) the application of the enhanced PC extraction step; bright fluorescent spots throughout the microparticle matrix depict the stained residual PC. The scale bar represents 50 µm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) showed a significant decrease of the glass transition temperature in untreated microparticles (8.2 °C) compared to PLGA microparticles with the enhanced PC extraction step (37.7 °C) (Fig. 7). Consequently, this clearly points to the high plasticizing effect of PC and also the importance of the enhanced PC extraction step for the obtainment of solid dry microparticles. Finally, microparticle stability has been analyzed in terms of a potential Tg shift of PLGA during storage at 5 °C ± 3 °C (Fig. 8A) and 25 °C ± 2 °C/60% RH ± 5% RH (Fig. 8B). Samples stored at 5 °C ± 3 °C for 12 months had a constant Tg value, and no significant Tg shift was observed. On the other hand, samples stored at 25 °C ± 2 °C/60% RH ± 5% RH had a constant Tg value (37.0 ± 0.8 °C) during the first 3 months, while after 6 months the Tg value gradually declined to 17.5 °C and was not observable after 12 months due to liquefaction. Samples stored at 40 °C ± 2 °C/75% RH ± 5% RH were showing signs of liquefaction after one week (data not shown) pointing at the instability of the samples at such conditions. Fig. 5. Hydrolytic treatment of (A) PC (empty squares), PLGA (full circles) and polysorbate 80 (empty circles) and comparison of their respective hydrolytic 4. Discussion profiles; (B) microparticle preparation mixtures evaluating the specific PLGA degradation products lactic and glycolic acid at the end of the production Polyester microparticles are regarded to be an attractive formula- procedure (mean ± SD, n = 3). tion approach in terms of biodegradability as well as biocompatibility (Anderson and Shive, 2012; Ignatius and Claes, 1996). Since they are typically used as a parenteral drug delivery formulation, low toxicity is a major requirement that expands to all involved excipients, including

217 ć D. Grizi , A. Lamprecht International Journal of Pharmaceutics 544 (2018) 213–221

organic solvents applied during the preparation step (Osterberg and See, 2003). In this context, distinct efforts have been made to replace standard organic solvents, for example with liquid polyethylene glycols (Ali and Lamprecht, 2013; Viehof et al., 2013). Despite having low toxic properties, polyethylene glycols show formulation issues such as high viscosity, slow PLGA solubility and potential stability issues due to the formation of peroxyl radicals (Gullapalli and Mazzitelli, 2015; Schou- Pedersen et al., 2014). In addition, it was reported that the nucleophilic side groups of low molecular polyethylene glycols (namely PEG 300) tend to form block-copolymers with PLGA (Schoenhammer et al., 2009). For this reason, the use of propylene carbonate as a non-toxic and partially water-miscible organic solvent can, on one hand, fulfill the safety requirements for such formulations and on the other hand provide more suitable physicochemical properties such as low viscosity or enhanced polymer solubility. A PC/water emulsion system was stabilized with polysorbate 80 forming an o/w emulsion as microparticle precursors similar to other conventional methods previously described (Elkharraz et al., 2011; Jeyanthi et al., 1996). The solvent extraction from the PLGA – rich droplet was initiated using distilled water as the extraction phase. In our case, this phenomenon was firstly tracked using nile red as a con- trast agent during drop to particle transformation. Since nile red exhibits very lipophilic properties, it will not leak from the inner polymer phase during PC diffusion (shown in the Video Supplement data). It was observed that the PC diffused out of the microparticle very fast (in form of a convective flow), leaving solidified porous particles with a rough surface. This clarified the observed roughness of the microparticle surface, which appeared during the drop to particle transformation and not as a result of drying. In addition, using water-soluble solvents (e.g. glycofurol or DMSO), may build porous microparticles due to the water intake (Allhenn and Lamprecht, 2011; Boimvaser et al., 2012). This porosity and the overall microparticle roughness could affect the degradation speed of the polymer matrix (Boimvaser et al., 2012) and finally have an impact on drug release kinetics. The high solubility of PLGA in PC is a major advantage of this method allowing for fast polymer solution preparation. However, it also represents one significant obstacle in view of its plasticizing effects which strongly affects the microparticle solidification. So far, different Fig. 6. Microparticle morphology observed after (A) 0.5 h, (B) 1.5 h and (C) studies pointed to such plasticizing influences of different organic sol- 4.5 h of the enhanced PC extraction step. The scale bar represents 100 µm. vents (Jain et al., 2000; Katou et al., 2008; Marquette et al., 2014).

Fig. 7. Impact of the enhanced PC extraction step on the PLGA Tg shift; lower curve: pure PLGA, middle curve: PLGA microparticles without the enhanced PC extraction step; upper curve: PLGA microparticles with the enhanced PC extraction step.

218 ć D. Grizi , A. Lamprecht International Journal of Pharmaceutics 544 (2018) 213–221

Fig. 8. Stability evaluation of PLGA microparticles at (A) 5 °C ± 3 °C and (B) 25 °C ± 2 °C/60% RH ± 5% RH during 12 months. All microparticle samples were prepared with the enhanced PC extraction step.

Precedent findings employing water-miscible organic solvents such as (Katou et al., 2008). This equilibrium can, however, be shifted by an glycofurol and ethyl acetate showed very pronounced plasticizing ef- increased solvent elimination from the extraction phase (Katou et al., fects, reflected in a significant Tg shift to lower temperatures (Allhenn 2008). Different from the existing approaches, i.e. evaporation of the and Lamprecht, 2011; Sah, 1997). solvent or further extraction by increased external phase volume, it was In the case where a classical solvent extraction step was applied achieved here by integrating an enhanced solvent extraction from the (without the enhanced PC extraction), a higher residual PC content was PLGA-rich droplet. The additional hydrolytic treatment triggers a PC impeding the formation of non-aggregated microparticles by its plas- mass transfer out of the particle matrix leading to residual PC con- ticizing effect. The resulting microparticle aggregation after drying was centrations far below typical values with similar solvents (Ali and also observed for other solvents such as ethyl acetate (Sah, 1997)or Lamprecht, 2013; Allhenn and Lamprecht, 2011; Viehof et al., 2013). ethyl formate (Sah, 2000) and could be prevented by lowering the re- The application of the hydrolytic treatment simultaneously enhances sidual solvent content (Marquette et al., 2014; Matsumoto et al., 2008; PC diffusion into the external phase and leads to the formation of Vay et al., 2012). propylene glycol. Moreover, this rapidly formed propylene glycol is a During the extraction step, the intraparticulate solvent content is in PLGA non-solvent, accelerating the PLGA solidification. It is also no- equilibrium with the solvent amount present in the extraction phase teworthy that the toxicity of propylene glycol is not an issue (McMartin,

219 ć D. Grizi , A. Lamprecht International Journal of Pharmaceutics 544 (2018) 213–221

2014). The instant PC degradation was observed during the drop to option of an enhanced PC extraction step enabled a viable preparation particle transformation investigation, where the surrounding of the method and solved the problems caused by high viscosity and residual droplets showed changes in light refraction when sodium hydroxide solvent content encountered with other non-toxic solvents. was present. The observed faster particle solidification in laser dif- Accordingly, the enhanced PC extraction step can be considered as a fraction measurements further confirmed the enhanced PC diffusion/ major advancement for the techniques employing low toxic cyclic ester- extraction out from the microparticle matrix. In addition, this analysis based solvents. The fact that no degradation issues were identified was performed after the emulsion formation during the PC extraction when other ester-based excipients were used underlines the robustness step. The apparent particle size increase for the sample without the of the method and suggests numerous pharmaceutical applications. enhanced PC extraction step is due to droplet collision and leads to a general increase in particle diameter (Fig. 4A) while the enhanced ex- Acknowledgments traction step leads to a more or less immediate solidification of the particle by building a polymer crust on the droplet surface. Accord- Daris Grizic would like to acknowledge the German Academic ingly, the particle diameter does not change during the entire pre- Exchange Service (DAAD) scholarship (91540177; A/13/91141). This paration period (Fig. 4B). work was partially supported by a French Government grant managed Even though the high hydrolytic reactivity of alkaline solutions by the French National Research Agency under the program towards PLGA (Croll et al., 2004) and polysorbate 80 (Kerwin, 2008)is “Investissements d’Avenir” with reference ANR-11-LABX-0021. well known, this study showed that the PC hydrolysis occurred faster and confirmed that all other excipients remained chemically un- Appendix A. Supplementary data affected. Even though acid-terminated PLGA (PLGA 502H) was used, which is known to be more hydrophilic and prone to degradation Supplementary data associated with this article can be found, in the compared to the end-capped PLGA (Ding and Schwendeman, 2004), no online version, at http://dx.doi.org/10.1016/j.ijpharm.2018.03.062. degradation was observed. The quantification of PLGA degradation products (namely lactic and glycolic acid) under microparticle pre- References paration conditions revealed that in the presence of PC, no PLGA degradation occurred and that polysorbate 80 also reduced the hydrolysis Ali, M.E., Lamprecht, A., 2013. Polyethylene glycol as an alternative polymer solvent for of the polymer. Besides, the overall duration of the enhanced PC ex- nanoparticle preparation. Int. J. Pharm. 456, 135–142. http://dx.doi.org/10.1016/j. ff ijpharm.2013.07.077. traction could be varied between 1.5 h and 4.5 h without a ecting Allhenn, D., Lamprecht, A., 2011. Microsphere preparation using the untoxic solvent microparticle morphology or PLGA degradation, which states clearly glycofurol. Pharm. Res. 28, 563–571. http://dx.doi.org/10.1007/s11095-010- the applicability and robustness of this methodology for the obtainment 0304-6. Anderson, J.M., Shive, M.S., 2012. Biodegradation and biocompatibility of PLA and PLGA of solid microparticles. microspheres. Adv. Drug Deliv. Rev. 64, 72–82. http://dx.doi.org/10.1016/j.addr. The fact that residual amounts of PC were still present in the matrix 2012.09.004. after the enhanced PC extraction was confirmed both by quantification Ao, Z., Li, Z., Zhang, G., Ngai, T., 2011. Colloidosomes formation by controlling the and visualization. It became clear that PC is localized inside small solvent extraction from particle-stabilized emulsions. Colloids Surf. A: Physicochem. Eng. Asp. 384, 592–596. http://dx.doi.org/10.1016/j.colsurfa.2011.05.011. droplet-shaped cavities inside the particle matrix, which were formed Beyer, K., Bergfeld, W., Berndt, W., Carlton, W., Hoffmann, D., Schroeter, A., Shank, R., during the microparticle solidification. A similar localisation pattern 1987. Final report on the safety assessment of propylene carbonate. Int. J. Toxicol. 6, – was reported from earlier studies when glycofurol was employed for 23 51. http://dx.doi.org/10.3109/10915818709095488. Bitz, C., Doelker, E., 1996. Influence of the preparation method on residual solvents in microparticle preparation (Allhenn and Lamprecht, 2011). However, biodegradable microspheres. Int. J. Pharm. 131, 171–181. http://dx.doi.org/10. the amounts present in the particle interior did not affect the me- 1016/0378-5173(95)04320-9. chanical stability of the dry microparticles. It should be mentioned Boimvaser, S., Cabrera, M.I., Grau, R., 2012. Degradation of porous implants formed in situ. Proc. Mater. Sci. 1, 454–460. http://dx.doi.org/10.1016/j.mspro.2012.06.061. however, that the presence of residual amounts of non-toxic solvents Clements, J.H., 2003. Reactive applications of cyclic alkylene carbonates. Ind. Eng. Chem. can also possess advantages such as solubility and dissolution rate im- Res. 42, 663–674. http://dx.doi.org/10.1021/ie020678i. provement, which was previously described for liquisolid formulations Croll, T.I., O’Connor, A.J., Stevens, G.W., Cooper-White, J.J., 2004. Controllable surface modification of poly (lactic-co-glycolic acid)(PLGA) by hydrolysis or aminolysis I: (Spireas and Sadu, 1998). physical, chemical, and theoretical aspects. Biomacromolecules 5, 463–473. In case residual solvents still remain present after microparticle Das, S., Mondal, A., Balasubramanian, S., 2017. Recent advances in modeling green solidification, storage stability evaluation was previously suggested solvents. Curr. Opin. Green Sustainable Chem. 5, 37–43. http://dx.doi.org/10.1016/ j.cogsc.2017.03.006. (Marquette et al., 2014). In this context, it was shown that residual Ding, A.M.Y.G., Schwendeman, S.P., 2004. In: Determination of Water-Soluble Acid ethyl acetate content of 36,400 ppm (3.64%) shifted the initial poly- Distribution in Poly (Lactide-co-glycolide), pp. 322–331. meric Tg from 53.9 °C to 37.2 °C. Interestingly, during storage at Dunne, M., Corrigan, O.I., Ramtoola, Z., 2000. Influence of particle size and dissolution 5 °C ± 3 °C, 25 °C ± 2 °C/60% RH and 40 °C ± 2 °C/75% RH for conditions on the degradation properties of polylactide-co-glycolide particles. Biomaterials 21, 1659–1668. http://dx.doi.org/10.1016/S0142-9612(00)00040-5. 12 weeks, the Tg shifted back to higher temperatures, which was at- Elkharraz, K., Ahmed, A.R., Dashevsky, A., Bodmeier, R., 2011. Encapsulation of water- tributed to the volatility of ethyl acetate and the subsequent migration soluble drugs by an o/o/o-solvent extraction microencapsulation method. Int. J. – and depletion during storage (Marquette et al., 2014). As PC is non- Pharm. 409, 89 95. http://dx.doi.org/10.1016/j.ijpharm.2011.02.029. Fujinaga, T., Izutsu, K., 1971. Propylene carbonate: purification and tests for purity. Pure volatile, the need for stability assessment in terms of a potential Tg shift Appl. Chem. 27, 273–280. http://dx.doi.org/10.1351/pac197127010273. is even higher. The observable microparticle liquefaction after twelve Grizić, D., Heimer, P., Vranić, E., Imhof, D., Lamprecht, A., 2016. Propylene carbonate quantification by its derivative 3,5-diacetyl-1,4-dihydro-2,6-lutidine. Talanta 151, months at 25 °C ± 2 °C/60% RH ± 5% RH is in accordance with – fi fi 75 82. http://dx.doi.org/10.1016/j.talanta.2016.01.022. previous ndings showing signi cant polymer degradation after Gullapalli, R.P., Mazzitelli, C.L., 2015. Polyethylene glycols in oral and parenteral for- 12 months at 25 °C (Dunne et al., 2000). Regardless of the present re- mulations – a critical review. Int. J. Pharm. 496, 219–239. http://dx.doi.org/10. sidual PC content in our study, short term storage (3 months) at 1016/j.ijpharm.2015.11.015. Herberger, J., Murphy, K., Munyakazi, L., Cordia, J., Westhaus, E., 2003. Carbon dioxide 25 °C ± 2 °C/60% RH ± 5% RH and long-term storage at 5 °C ± 3 °C extraction of residual solvents in poly(lactide-co-glycolide) microparticles. J. Control. (> 12 months) showed acceptable stability profiles underlining the Release 90, 181–195. http://dx.doi.org/10.1016/S0168-3659(03)00152-4. feasibility of the microparticle formulation approach. ICH, 2016. ICH Guideline Q3C (R6) on Impurities: Guideline for Residual Solvents. Ignatius, A.A., Claes, L.E., 1996. In vitro biocompatibility of bioresorbable polymers: poly (L, DL-lactide) and poly (L-lactide-co-glycolide). Biomaterials 17, 831–839. 5. Conclusion Jain, R.A., Rhodes, C.T., Railkar, A.M., Malick, A.W., Shah, N.H., 2000. Controlled release of drugs from injectable in situ formed biodegradable PLGA microspheres: effect of – PC revealed excellent properties as a low toxic organic solvent for various formulation variables. Eur. J. Pharm. Biopharm. 50, 257 262. http://dx.doi. org/10.1016/S0939-6411(00)00062-X. polyester microparticle preparation. Especially, the unique technical

220 ć D. Grizi , A. Lamprecht International Journal of Pharmaceutics 544 (2018) 213–221

Jeyanthi, R., Thanoo, B.C., Metha, R.C., DeLuca, P.P., 1996. Effect of solvent removal Sah, H., 2000. Ethyl formate – alternative dispersed solvent useful in preparing PLGA technique on the matrix characteristics of polylactide/glycolide microspheres for microspheres. Int. J. Pharm. 195, 103–113. http://dx.doi.org/10.1016/S0378- peptide delivery. J. Control. Release 38, 235–244. http://dx.doi.org/10.1016/0168- 5173(99)00379-8. 3659(95)00125-5. Sah, H., 1997. Microencapsulation techniques using ethyl acetate as a dispersed solvent: Katou, H., Wandrey, A.J., Gander, B., 2008. Kinetics of solvent extraction/evaporation effects of its extraction rate on the characteristics of PLGA microspheres. J. Control. process for PLGA microparticle fabrication. Int. J. Pharm. 364, 45–53. http://dx.doi. Release 47, 233–245. http://dx.doi.org/10.1016/S0168-3659(97)01647-7. org/10.1016/j.ijpharm.2008.08.015. Schoenhammer, K., Petersen, H., Guethlein, F., Goepferich, A., 2009. Injectable in situ Kerwin, B.A., 2008. Polysorbates 20 and 80 used in the formulation of protein biother- forming depot systems: PEG-DAE as novel solvent for improved PLGA storage sta- apeutics: structure and degradation pathways. J. Pharm. Sci. 97, 2924–2935. http:// bility. Int. J. Pharm. 371, 33–39. http://dx.doi.org/10.1016/j.ijpharm.2008.12.019. dx.doi.org/10.1002/jps.21190. Schou-Pedersen, A.M.V., Hansen, S.H., Moesgaard, B., Østergaard, J., 2014. Kinetics of Kim, Y., Kim, H., Sah, H., 2016. Methylamine acts as excellent chemical trigger to Harden the esterification of active pharmaceutical ingredients containing carboxylic acid emulsion droplets into spongy PLGA microspheres. RSC Adv. 6, 85275–85284. functionality in polyethylene glycol: formulation implications. J. Pharm. Sci. 103, http://dx.doi.org/10.1039/C6RA17576G. 2424–2433. http://dx.doi.org/10.1002/jps.24062. Li, J., Rothstein, S.N., Little, S.R., Edenborn, H.M., Meyer, T.Y., 2012. The effect of Shaikh, A.-A.G., Sivaram, S., 1996. Organic carbonates. Chem. Rev. 96, 951–976. http:// monomer order on the hydrolysis of biodegradable poly(lactic-co-glycolic acid) re- dx.doi.org/10.1021/cr950067i. peating sequence copolymers. J. Am. Chem. Soc. 134, 16352–16359. http://dx.doi. Shao, T., Bai, L., Yan, B., Jin, Y., Cheng, Y., 2017. Modeling the solidification of O/W- org/10.1021/ja306866w. emulsion droplet in solvent evaporation technique. Chem. Eng. Res. Des. 122, Marquette, S., Peerboom, C., Yates, A., Denis, L., Langer, I., Amighi, K., Goole, J., 2014. 233–242. http://dx.doi.org/10.1016/j.cherd.2017.04.022. Stability study of full-length antibody (anti-TNF alpha) loaded PLGA microspheres. Sommer, L., Ackermann, G., Burns, D.T., 1990. Present and future status of organic Int. J. Pharm. 470, 41–50. http://dx.doi.org/10.1016/j.ijpharm.2014.04.063. analytical reagents - Part II. Inorganic chemical analysis: classical methods, molecular Matsumoto, A., Kitazawa, T., Murata, J., Horikiri, Y., Yamahara, H., 2008. A novel pre- spectroscopy (absorption and emission) and solvent extraction. Pure Appl. Chem. 62, paration method for PLGA microspheres using non-halogenated solvents. J. Control. 2323–2338. http://dx.doi.org/10.1351/pac199062122323. Release 129, 223–227. http://dx.doi.org/10.1016/j.jconrel.2008.04.008. Song, K.C., Lee, H.S., Choung, I.Y., Cho, K.I., Ahn, Y., Choi, E.J., 2006. The effect of type McMartin, K., 2014. Propylene glycol. In: Encyclopedia of Toxicology. Elsevier, pp. of organic phase solvents on the particle size of poly(D, L-lactide-co-glycolide) na- 1113–1116. http://dx.doi.org/10.1016/B978-0-12-386454-3.01029-0. noparticles. Colloids Surf. A: Physicochem. Eng. Asp. 276, 162–167. http://dx.doi. Osterberg, R.E., See, N.A., 2003. Toxicity of excipients – a food and drug administration org/10.1016/j.colsurfa.2005.10.064. perspective. Africa (Lond) 377–380. http://dx.doi.org/10.1080/ Spireas, S., Sadu, S., 1998. Enhancement of prednisolone dissolution properties using 10915810390232330. liquisolid compacts. Int. J. Pharm. 166, 177–188. http://dx.doi.org/10.1016/S0378- Quintanar-Guerrero, D., Fessi, H., Allémann, E., Doelker, E., 1996. Influence of stabilizing 5173(98)00046-5. agents and preparative variables on the formation of poly(D,L-lactic acid) nano- Vay, K., Frieß, W., Scheler, S., 2012. A detailed view of microparticle formation by in- particles by an emulsification-diffusion technique. Int. J. Pharm. 143, 133–141. process monitoring of the glass transition temperature. Eur. J. Pharm. Biopharm. 81, http://dx.doi.org/10.1016/S0378-5173(96)04697-2. 399–408. http://dx.doi.org/10.1016/j.ejpb.2012.02.019. Raymond, C.R., Paul, J.S., Fuinn, M.E.Q., 2009. In: Polyoxyethylene Sorbitane Fatty Acid Viehof, A., Javot, L., Béduneau, A., Pellequer, Y., Lamprecht, A., 2013. Oral insulin de- Esters. Handb. Pharm. Excipients, sixth ed. pp. 549–553. http://dx.doi.org/10.1016/ livery in rats by nanoparticles prepared with non-toxic solvents. Int. J. Pharm. 443, S0168-3659(01)00243-7. 169–174. http://dx.doi.org/10.1016/j.ijpharm.2013.01.017. Rosca, I.D., Watari, F., Uo, M., 2004. Microparticle formation and its mechanism in single Yang, Y.L., Ramaswamy, S.G., Jakoby, W.B., 1998. Enzymatic hydrolysis of organic cyclic and double emulsion solvent evaporation. J. Control. Release 99, 271–280. http://dx. carbonates. J. Biol. Chem. 273, 7814–7817. http://dx.doi.org/10.1074/jbc.273.14. doi.org/10.1016/j.jconrel.2004.07.007. 7814.

221